On Monday 26 September 2011 19:32:47 Mauro Carvalho Chehab wrote: > Em 26-09-2011 05:13, Thomas Petazzoni escreveu: > > Le Fri, 23 Sep 2011 23:15:54 -0300, Mauro Carvalho Chehab a écrit: > >>> And still the result is the same: we get a first frame, and then > >>> nothing more, and we have a large number of error messages in the > >>> kernel logs. > >> > >> I don't think that this is related to the power manager anymore. It can > >> be related to cache coherency and/or to iommu support. > > > > As you suspected, increasing PWR_SLEEP_INTERVAL didn't change anything. > > What do you suggest to track down the potential cache coherency issues ? Are you using the MMAP or USERPTR capture method ? If using MMAP, can you try (as a test only) to unmap the buffer before queueing it and to remap it after dequeuing it ? > Take a look at the ML. The SoC people discussed a lot about cache > coherency problems and how to solve it. Videobuf2 has a better support > on embedded world. I would take a look on it and see what it does different > than other drivers. Maybe Jonathan Corbet patches for the ccic driver may > help you. > > It is probably a good idea to change cx231xx to use videobuf2, in order to > fix this issue. -- Regards, Laurent Pinchart -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
